> It's not perfect, but the distance from your house to a ShotSpotter sensor correlates fairly well with your household income. The wealthier you are, the less surveilled you are.
To give ABQ police the benefit of the doubt, that pattern could also be compatible with more gun crime equaling more surveillance. It would be nice to have enough gun crime and sensor location data to see how true that is. When the sensors are as dense as they are, it's not clear that knowing the sensor locations is an advantage to offenders, at least in the gunshot spotting role.
If anyone wants the raw data, it's available in window._Flourish_data variable on https://flo.uri.sh/visualisation/16818696/embed
Which means you can extract it with my https://shot-scraper.datasette.io/ tool like this:
shot-scraper javascript \
'https://flo.uri.sh/visualisation/16818696/embed' _Flourish_data \
> /tmp/data.json
I loaded it into SQLite like this:
cat /tmp/data.json | jq .events | \
sqlite-utils insert /tmp/shots.db locations -
Timely, as a boy was shot at last week in Chicago while setting off fireworks that triggered a ShotSpotter alert:
https://chicago.suntimes.com/crime/2024/02/27/chicago-police...
What are the potential downsides to police oversight?
> But, if asked, they provide a form letter written by ShotSpotter. Their contract prohibits the disclosure of any actual data.
A potentially dystopian surveillance apparatus is installed citywide, and the police can't discuss about it's efficiency because the company that sold it to them won't let them?
I would request that a crowd as technically sophisticated as HN move past marketing claims when discussing this system. "Gunshot detection" is pure marketing.
The devices are hot mics — public audio surveillance. The software layer triggers alerts on loud noises, which are sent off to a facility that is little more than a call center, for characterization by a human worker there. This system detects door slams and popped volleyballs just as well as it detects gun fire (which is to say, imperfectly on all counts).
I had no idea ShotSpotter was recording conversations, always assumed the nature of analyzing for a sound signature would be throwing much of the data away immediately
i broadly agree with the sentiment in this article, especially the "no public oversight" part.
but this:
> The reader can probably infer how this coverage pattern relates to race and class in Albuquerque. It's not perfect, but the distance from your house to a ShotSpotter sensor correlates fairly well with your household income.
is facetious and is almost certainly argued in bad faith. well, duh. the distance to the nearest "shot spotter" box also correlates with the incidence of crime and gunfire in the area. to bring up racism or classism is unhelpful. that correlation is unfortunate, likely true, and also not the problem at hand.
I've also been blogging about ShotSpotter and review the evidence that the system is good or bad.
https://quickthoughts.substack.com/p/shotspotter-good-or-bad
There's a lot of evidence that ShotSpotter detects almost all gunshots and that's been validated by multiple third party groups. ShotSpotter also seems to alert to things that are either false positives (construction sounds, fireworks, etc) or are not useful.
Chicago IG says 9/10 times when officers respond to a ShotSpotter alert they find nothing. 1/10 times it leads to an arrest, but the arrest isn't always strictly related to the ShotSpotter event - e.g. police responding to an event stop a speeding car and discover drug paraphernalia or an unrelated gun.
Some cities, e.g. Atlanta, discontinued ShotSpotter over cost benefit concerns. From their analysis it seemed a better use of money to hire more officers than use ShotSpotter. Still, it's in use in 84 metro areas today.
Ultimately, I think it should be up to the local community. The individual community is the one who will most benefit (if it is beneficial) and suffer from increased police incursions.
In Dallas I hear the pop-pop-pop of gunfire virtually every night in my oakcliff neighborhood. Even when the Cowboys score a touchdown you hear gunfire. The police have basically given up.
I can't help but think how easily such a system - which is hardly sophisticated, in technical principle - could be used to pinpoint things that are 'problematic' in my neighbourhood. Loutish vehicle behaviour of various kinds. The loud Harley Davidson and dirt-bike signatures associated with the comings-and-goings of local gang members. Squealing tyres of deliberate burn-outs on the road. The occasional fool loudly blasting down the road at twice the speed limit. Things which the current system of "wait until someone gets annoyed enough or gets the courage to call the police to complain, and one might arrive 15 minutes later, by which time the coop has been long flown" allow to run mostly unchecked.
But those thoughts go hand in hand with a vague but distinct discomfort.
>Many assumed that ShotSpotter coverage was concentrated in disadvantaged parts of the city, an unsurprising outcome but one that could contribute to systemic overpolicing.
The author presents this as a negative but it is obviously a good thing. If there was an increase in gunfire in my neighbourhood I would hope that police increases their presence, lest the "disadvantages" begin to accumulate.
It just makes sense to put the sensors near the phenomena being sensed, and if people know where they are they could be manipulated. If guns are going off and people are getting caught, that seems like effective policing, not over or under policing.
Several of us are noting the less than 5% success rate in finding definitive evidence of shots fired, and calling the tech a failure.
I disagree, because of the importance of preventing active shooters. With this kind of system you are going to want to lean toward false positives vs false negatives.
There is definitely a lot of room for improvement though. Improvement could come through better ML use/training, better sensors, and better human in the loop analysts.
I am firmly with Benjamin Franklin, in that "Those who would give up essential Liberty, to purchase a little temporary Safety, deserve neither Liberty nor Safety.". But Franklin himself was in favor of collective security, and Franklin said this in support of the state's authority to provide collective security ( https://www.npr.org/2015/03/02/390245038/ben-franklins-famou... ). Also, these sensors monitor public spaces, where no expectation of privacy exists.
So I am for more sensors, video cameras on the sensors, more ML on the sensor data, and better training for analysts, but I am also for complete transparency on where sensors are and how they are used, with a publicly released report from annual external party review of police conduct in the use or abuse of sensor data.
Interesting how takes on ShotSpotter vary depending on the context and situation. In San Jose it was used to help enforce rules against firing guns in the air. In these cases there would usually be reports to police, but finding the shooter was extremely difficult. With ShotSpotter it became possible to identify locations on individual lots. Using this information enforcement became much more effective and there was a huge reduction in celebratory gunfire. In a large number of these cases no charges needed to be filed as gun owners were often uninformed, apologetic, and compliant. From the start of this program it was explicitly stated that ShotSpotter information would only be used by constables on patrol and would not be admitted as evidence in court. System coverage was based on where crimes were being reported. That such criteria overlap with race and class makes sense without necessarily being connected.
It would be interesting to know if there might be some reasonable bounds that could be used to enable ShotSpotter to be used without being considered intrusive surveillance. Having information about gunshots or possible gunshots can be extremely useful for responders who need to understand where events are happening. This does not necessarily mean that the system has to be open to other uses, especially recording and replay of audio.
I have a project for the Raspberry Pi that provides for sound localization via time difference of arrival (TDOA). In a similar manner I suppose as shotspotter.
There was a case recently in America where a grandfather was jailed for a significant part of a year based on a shotspotter localization and no other physical evidence such as gunshot residue.
If citizens in such areas run their own systems they would have a means to provide counter evidence to that provided by shotspotter. Currently they have no means to do that. Even the times of the shots they have to take shot spotters word for it.
Recordings by citizens themselves is an inherently safer approach because their sphere of influence is considerably smaller. Recordings are written to a separate partition on an SD card, it’s pretty simple to encrypt this partition as well if you like, I’ve done that.
For those who are interested each node can be made very cheaply. It will run on a Raspberry Pi zero with a 7 euro GPS. It can also run portably on batteries.
Here are relevant links:
https://github.com/hcfman/sbts-aru
https://hackaday.com/2023/12/30/localizing-fireworks-launche...
https://medium.com/@kim_94237/tdoa-sound-localization-with-t...
Hard to imagine this can discern between things like engine backfires, fireworks and garbage trucks effectively enough to be anything more than a really expensive checkmark on some captains yearly performance review.
One thing missed in this writing is that shotspotter also saves lives. In my city, people shot in the alley and left for dead are found shortly after they are shot because of the detection.
Without this tech, some would be left to a slow death in an alley where no one would find them until morning.
Here’s my ask before you enter your perfectly argued comment here: if you have 30 mins read these NYT profiles of 12 children lost to gun crime: https://www.nytimes.com/interactive/2022/12/14/magazine/gun-....
If you’re passionate about this subject and have lots of time go and volunteer at high crime communities (if you dare), for example in high schools.
I did the latter for three years, volunteering to help the Ace Tech High School (https://en.m.wikipedia.org/wiki/ACE_Amandla_Charter_High_Sch...) more than 10 years ago, as part of their Robotics Team. The problems I saw were myriad but it was common to see lockers decorated with flowers, for kids who died recently.
I totally understand and agree with the educated person’s poverty, bias, etc. arguments. But when a problem gets this bad first you need to stop it, and then work on long term goals.
Just answer this question: if your child had 10% chance of getting shot and killed walking to/from from school, what would your position be on any technology that can reduce that by even 1%.
Spoiler alert: with a secondary audio source you can map out physical spaces as easily as one might with lidar
I have some noisy neighbours and I've thought of the same idea for controlling noise in neighbourhoods. You could have a microphone on every lamp post and send people fines for violating the rules.
This is a strange conversation, because surely people realise that all CCTV cameras made in the past 10 years have microphones and record audio? Audio street surveillance is ubiquitous.
I think it's lame to limit the detection to just gun shots. Like the web has analytics, I think real life could have audiolytics. One example: I can collect all the data I want from my backyard and give a report to my neighbor about exactly how long and how loudly their dog has been barking.
I can hear which birds have been in my backyard.
I can... hear what a customer says about my product in-store and offer them coupons at checkout or later or get feedback about my product...
They are probably solar powered running on something with low power so they don't have to restrict the locations of hiding these devices to areas directly connected to the electrical grid.
Something along these same lines — I feel like I’m seeing those “Flock” cameras everywhere now. So between Shotspotter and Flock, you’re pretty much always under surveillance.
instead of fixing the reasons these shootings happen, usa as usual tries to ameliorate the consequences with more surveillance/technology
Do these record audio constantly or only after shootings?
Seems like societally we would want audio to be recorded after shootings?
It's used the way police use any technology: to over-police the poor and non-white communities. https://www.cnn.com/2024/02/24/us/shotspotter-cities-choose-...
I don’t understand the pearl clutching over what parts of town these sensors get concentrated in. Wouldn’t it just be wasteful to put them in areas where people don’t have the habit of shooting each other? Is it really overpolicing when those who do are made to face the consequences?
I've experienced living in a neighborhood covered by shot spotter and hearing gun shots outside, and I think shot spotter is great. I want the cops to actually show up and arrest whoever is terrorizing everybody. Thank you very much
I had no idea. Two key takeaways...
> Conversations recorded by ShotSpotter sensors have twice been introduced as evidence in criminal trials. In one case the court allowed it, in another the court did not.
> ShotSpotter sensor correlates fairly well with your household income. The wealthier you are, the less surveilled you are.
[dead]
Wait what this is real? I thought it was just a plot point invented for Person of Interest
[flagged]
A few year down the road the Shirky principle demands a revelation that ShorSpotter sponsored the NRA.
It’s amazing lime 500 years later, there’s still a developed country on earth that hasn’t figured the externalities of gunpowder
[flagged]
[flagged]
[flagged]
[flagged]
[dead]
Sounds like it might be a good idea to add a sound maker near these devices.
Things like ShotSpotter are only fighting the symptoms. As an european I find it ridiculus to hear that such thing even exists. Why would we not simply ban the guns themselves?
This phrase "systemic overpolicing" and ShotSpotter both come from the same mind - the mind of the state.
The state - which like most organizations is concerned mostly with preserving itself - has an interest in the surveillance of the population and in shifting blame from it's failures to deal with poverty and crime to skapegoats: the police, racism, property taxes, etc.
The west is run by priests (professors, advisors, journalists, students, diplomats) with the support of the merchants. Priests always pretend it is flipped, but it's not. One "tell" is that the priests are never the villains in Hollywood movies. The other groups (warriors, merchants, and peasants) all do even villain duty.
Bias shot-spotter placement is a classic case of priests blaming merchants. There might even be something to the substance, but the priests run the show - not SoundThinking Inc.
Hello all, occasionally I write what I consider "Albuquerque content" and I do not expect it to become broadly popular. This article is something I put together very quickly and it probably assumes a certain degree of familiarity with the political context around policing in Albuquerque (which either side will tell you is very contentious) and, more broadly, policing and civil justice. Even without the currently evolving bribery scandal, the level of public trust in APD (and even the city council's confidence in APD) is very low. APD's transparency and accountability, or lack thereof, has been a common locus of the debate. On the other hand, another major issue has been APD's chronic understaffing, and APD contends that ShotSpotter and other elements of their real-time program help to close the gap that results from their limited personnel. With gun crime as one of the foremost issues in the city, whether you view APD positively or negatively, ShotSpotter is a big part of the discussion right now.
Historically, APD's use of pervasive surveillance technology has been a flashpoint in the debate. APD has live access to perhaps 3-4 thousand cameras across the city (they aren't very transparent about this and it depends on how far along the APS integration project is), they have used facial recognition against driver's license photos and other sources since 2014, they have installed ALPR throughout the city and recently expanded retention to one year, etc. This is all fed into the Real-Time Crime Center, which uses a data fusion product from a vendor called Genetec to provide sort of a futuristic point-and-click data system that combines ShotSpotter detections with video feeds with service call records etc. to produce sort of a dossier on any given person or location.
Unfortunately, there are a lot of things going on in city politics, especially with regard to crime and policing, and so the topic of surveillance has mostly fallen out of public attention.
Still, APD's refusal to say in any detail what parts of the city were covered by ShotSpotter has been one of the big ongoing frustrations, particularly among those who favor police reform. I mostly wrote this article to highlight that there is finally information on the matter available. The concerns about how distribution of sensors and, more broadly, use of surveillance technology impacts civil rights and quality of life in the city are mentioned mainly as an aside and I do not attempt to articulate the pros and cons. That would require a rather lengthy piece as the topic is complex, and currently the greater part of the controversy isn't even about the wisdom of deploying ShotSpotter, but rather over whether or not ShotSpotter even works (and, consequently, whether or not it's simply a waste of city money, at a rate of around $5 million).